1. Conservation Agriculture Adoption by Cotton Farmers in Eastern Zambia Philip Grabowski, John Kerr, Steve Haggblade and Stephen Kabwe

2. Cotton is a key cash crop Contract farming Equipment and herbicides available on credit Private sector extension

3. Conservation agriculture Minimum tillage and mulching early planting reduce evaporation increase infiltration of water Rotations, manure application and precise fertilizer improved nutrient cycling

4. Adoption of CA by smallholder cotton farmers in Zambia Percent of NWK farmers using CF on cotton in Eastern Province (Grabowski et al. forthcoming)

5. Spatial variation in adoption levels Percent of NWK and Cargill farmers using CF on cotton by ward in Eastern Province (Grabowski et al. forthcoming)

6. Research Question What are the factors driving farmer decision making about their land preparation methods? Hypotheses Labor, wealth, input availability, promotion, experience and soil characteristics affect the probability of CA use

7. Methods – Survey and Interviews 326 Farmers from NWK and Cargill 775 Plots of cotton and maize 12 Plot level variables 13 Household level variables 6 Community level variables

8. Opinions comparing basins with hoeing and ripping with plowing 80% or more said basins and ripping: ◦ have better yields ◦ do better on dry years ◦ are better for the soil and have less erosion Results were divided on if CA or conventional was less work or had less weeds Over 50% said conventional did better than CA on wet years

9. Land preparation by crop

10. The reality of CA use Only 5% of basin plots and 13% of ripped plots had a legume the previous season 96% of basin plots and 62% of ripped plots had the residues heavily grazed 85% of ripped plots were “banked” – tilled to control weeds and reduce lodging 52% of basin plots were in their first year in that location – often rotated with maize

11. Multinomial logit model Estimate how each variable affects the probability that a plot will be prepared using any one of these four methods: 1. Plowing (P) 2. Ripping (R) 3. Hoeing (H) 4. Basins (B)

12. Abbreviated Regression Results * is for p < 0.1, ** for p < 0.05, and *** for p < 0.01 Explanatory VariablesPlow vs. HoeRipping vs. PlowBasin vs. Hoe Plot Level Plot area (ha)0.39 -0.11 -2.34* Flat (Y/N)-0.07 0.42 2.33*** RankRatio (low = good)0.58 0.87 3.18*** Tenure Secure (Y/N)-0.15 0.54 -2.24** Household Level Trained in CA (Y/N)1.96***3.89***2.16** Workers per hectare1.02**-1.07 1.03 Wealth index5.27***0.98*1.67 Cotton experience (Yrs)0.04 0.13***0.13** Community Level Years CA promoted-0.50**-0.15 -0.56** Buyer CA practice0.61 0.91***0.36 Herbicides avail. (Y/N)5.56**-0.31 -1.15 Pop. density (1000/km 2 )-1.84 15.7 -25.4***

13. Key plot-level variables

14. Key household-level variables

15. Key community-level variables

16. Summary – how CA is used Basins tend to be used by labor abundant and land constrained households to make marginal land more productive for maize. Ripping tends to be used by wealthier farmers with oxen who can buy the equipment. Rental markets have not developed as expected because of concerns for oxen health.

17. Implications Basins meet a niche need but are unlikely to be adopted on a large scale Ripping has higher adoption potential especially if rental availability is increased ◦ Ownership requires investments in equipment and animals ◦ Rental markets require healthier oxen ◦ Availability must be accompanied by training

18. Thank you! Any questions? grabow21@msu.edu